From brain to screen: a scientist is ready to directly turn what you think into words
- A new method created by a team of Stanford University researchers allows a patient to type with record-breaking speed using only his brain
- Brain-computer interface technology has gained public attention after Elon Musk’s Neuralink released video of a monkey playing video games with its mind
A breakthrough in brain-computer interface (BCI) has enabled a man with paralysed hands to type 90 characters a minute just by thinking about the words, making him the world’s fastest mind-typist using that brain-reading technology, according to researchers from Stanford University.
Their study, published in Nature in May and presented at WE Summit, a science conference hosted by Tencent Holdings online on Saturday, enabled the patient to achieve a typing speed twice as fast as people who relied on an older type of BCI technology, which let users type by using their mind to move a cursor to the desired character on a virtual keyboard.
“That’s just the tip of the iceberg. When we really understand the brain through neuroscience in the coming decades we should be able to do much better in a wider variety of tasks,” neuroscientist and engineer Krishna Shenoy said at the summit.
In Shenoy’s experiment, researchers implanted electrode sensors in the patient’s brain, which recorded the electrical activity of neurons as the patient imagined himself writing with his hand. An artificial intelligence (AI) algorithm was used to decode the signals and display the letters on a computer screen.
The new method, used in conjunction with an AI-powered language system with an autocorrect function, boosted the accuracy rate to 99 per cent, researchers said.
While Shenoy said that 15 years of experiments by his research team has proved that the surgical implantation method they use is safe, public perception will continue to be one of the major roadblocks to mass market adoption of BCI technology.
To make commercialisation possible, scientists will also need to safely insert more electrodes in the brain. The brain has about 86 billion neurons, but scientists are currently only able to record signals from less than a few thousand of them, Shenoy said.
Still, the biggest challenge would be to find a viable business model for the technology, since neurosurgery is expensive.